Measuring cell-to-cell expression variability in single-cell RNA-sequencing data: a comparative analysis and...

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) captures gene expression heterogeneity.
• Quantifying cell-to-cell variability is crucial but challenging due to scRNA-seq data structures like zero inflation.

Purpose of the Study:

• To systematically evaluate 14 different metrics for measuring cell-to-cell variability in transcriptomic data.
• To identify the optimal statistical approach for analyzing scRNA-seq data, considering data-specific features and biological properties.

Main Methods:

• Systematic evaluation of 14 variability metrics using simulations and real datasets.
• Benchmarking metric performance against data features (sparsity, platform) and biological variability.
• Application of the top-performing metric (scran) to analyze B cell differentiation and aging.

Main Results:

• The scran metric showed the strongest all-round performance in quantifying cell-to-cell variability.
• Analysis revealed unique gene signatures with distinct expression profiles during B cell differentiation.
• Differentially variable genes between young and old cells identified regulatory changes potentially missed by mean expression analysis.

Conclusions:

• Capturing cell-to-cell gene expression variability is vital for understanding complex biological processes like differentiation and aging.
• The findings emphasize the value of these methods for analyzing individual cell types and uncovering regulatory insights.